A Systems-Approach Framework for Fatigue-Aware Interactive Evolutionary Design Adaptive Human-AI Evaluation and Generative Co-Evolution

This study presents a fatigue-aware interactive evolutionary art design framework conceptualized as a sociotechnical information system in which users, algorithms, evaluation signals, and generative models jointly influence design outcomes. Existing approaches commonly assume stable user evaluation, neglecting fatigue-induced variations in cognitive load that undermine evaluation reliability and system performance. To address this limitation, this paper proposes an adaptive interactive genetic algorithm integrating a dynamically estimated fatigue index. User behavioral patterns and lightweight physiological signals are modeled via a temporal neural network to capture fatigue evolution. The resulting index is embedded into fitness evaluation, population scheduling, interaction pacing, and surrogate-assisted automatic evolution. A generative adversarial network is also incorporated to expand the design space and coevolve with genetic operators.